Font Size: a A A

Design And Implementation Of Financial Field Hot Topic Detection And Analysis Based On Microblog

Posted on:2017-05-03Degree:MasterType:Thesis
Country:ChinaCandidate:E Z HuaFull Text:PDF
GTID:2348330518495348Subject:Computer technology
Abstract/Summary:PDF Full Text Request
Microblog is a social platform of social entertainment,news sources,information dissemination,which has a large user group.The user scale of online stock trading and financial transactions has increased dramatically.Weibo produce large amounts of information data every day.These information involves many industries and a wide range of coverage.Timeliness and high authority of these information,is an important reason for investors and financial managers of particular concern.How to find financial hot topics from a large number of Sina weibo data,which has become securities company and finance company's central issue.This paper is mainly to solve the above problem that extracts financial hot topics from Sina weibo.This paper firstly introduces the related technology of topic detection,and the related technology of clustering algorithm.Then we analyze the clustering algorithm,choose the Single-Pass algorithm as the text clustering algorithm and propose the improved algorithm.In order to improve the IDF of TF-IDF is a constant value,not with the data set dynamic problem,we propose incremental TF-IWF-IDF of feature items position weight calculation method.The traditional feature vectors ignore the semantic and context of feature items.Therefore,in this paper,a new feature vector representation method is proposed,which is called an incremental TF-IWF-IDF based on Word2Vec method.This paper proposes Two Steps of Single-Pass based on Multi Topic Centers method to solve the problems of Single-Pass algorithm.For weibo dataset,by experimental comparison,the improved algorithm has better performance than unimproved Single-Pass algorithm by about 10%in hot topic detection and tracking of this paper.Finally,this paper designs and implements the financial hot topics prototype system based on improved clustering algorithm.Based on the analysis of the functional requirements,this paper gives the design and implementation of the system architecture and function module,and shows the prototype system renderings.
Keywords/Search Tags:TF-IWF-IDF, Word2Vec, Single-Pass, Topic Detection and Tracking
PDF Full Text Request
Related items